NSP4 and ORF9b of SARS-CoV-2 Induce Pro-Inflammatory Mitochondrial DNA Release in Inner Membrane-Derived Vesicles

Circulating cell-free mitochondrial DNA (cf-mtDNA) has been found in the plasma of severely ill COVID-19 patients and is now known as a strong predictor of mortality. However, the underlying mechanism of mtDNA release is unexplored. Here, we show a novel mechanism of SARS-CoV-2-mediated pro-inflammatory/pro-apoptotic mtDNA release and a rational therapeutic stem cell-based approach to mitigate these effects. We systematically screened the effects of 29 SARS-CoV-2 proteins on mitochondrial damage and cell death and found that NSP4 and ORF9b caused extensive mitochondrial structural changes, outer membrane macropore formation, and the release of inner membrane vesicles loaded with mtDNA. The macropore-forming ability of NSP4 was mediated through its interaction with BCL2 antagonist/killer (BAK), whereas ORF9b was found to inhibit the anti-apoptotic member of the BCL2 family protein myeloid cell leukemia-1 (MCL1) and induce inner membrane vesicle formation containing mtDNA. Knockdown of BAK and/or overexpression of MCL1 significantly reversed SARS-CoV-2-mediated mitochondrial damage. Therapeutically, we engineered human mesenchymal stem cells (MSCs) with a simultaneous knockdown of BAK and overexpression of MCL1 (MSCshBAK+MCL1) and named these cells IMAT-MSCs (intercellular mitochondrial transfer-assisted therapeutic MSCs). Upon co-culture with SARS-CoV-2-infected or NSP4/ORF9b-transduced airway epithelial cells, IMAT-MSCs displayed functional intercellular mitochondrial transfer (IMT) via tunneling nanotubes (TNTs). The mitochondrial donation by IMAT-MSCs attenuated the pro-inflammatory and pro-apoptotic mtDNA release from co-cultured epithelial cells. Our findings thus provide a new mechanistic basis for SARS-CoV-2-induced cell death and a novel therapeutic approach to engineering MSCs for the treatment of COVID-19.


Introduction
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) can directly or indirectly induce airway epithelial cell death [1][2][3]. In direct virus-induced cell death, viral proteins hijack the host cell's anti-apoptotic proteins and subsequently activate pro-apoptotic through which mammalian cells, especially MSCs, exert their therapeutic effects. In this study, we found that SARS-CoV-2 and its encoded proteins NSP4 and ORF9b synergistically induced outer membrane macropore formation and inner membrane vesicle formation to facilitate extracellular mtDNA release. The released mtDNA induced a pro-inflammatory response in the target airway epithelial cells, leading to cell death. To design effective therapeutic strategies, MSCs were genetically engineered to harbor macropore-deficient mitochondria by simultaneously downregulating BAK and overexpressing MCL1. The donation of the engineered mitochondria by MSCs to SARS-CoV-2-infected or NSP4-ORF9b-expressing airway epithelial cells mitigated SARS-CoV-2-induced pro-inflammatory mtDNA release and rescued airway epithelial cell death.

Ethical Clearance and Patient Information
Blood samples from the patients used in the present study were obtained from the Safdarjung Hospital. The blood samples were collected in vacutainers containing ethylenediaminetetraacetic acid (EDTA) (POLYMED Medical Devices, India). Human ethical clearance (IEC/VMMC/SJH/Project/2020-07/CC-06) was obtained for blood collection. Further, institutional ethical clearance (1/10/290/JMI/IEC/2020) and biosafety clearance (Ref. No. P1/12-21.12.2020) were also obtained from Jamia Millia Islamia as mentioned by us previously [41]. Patient consent was obtained while collecting the samples according to the ICMR GCP guidelines.

Sample Collection and Processing
In total, 162 blood samples were collected for the study by a trained clinician from Safdarjung hospital, New Delhi. Samples collected between January 2022 and March 2022 are referred to as OmicronS (Omicron-suspected). Similarly, as described by us previously [41], samples obtained between June 2020 and October 2020 are referred to as COVID-19. The collected samples included 89 COVID-19-positive, 35 COVID-19negative, 14 convalescent patients (recovered with confirmed negative RT-qPCR results), and 24 new cases of OmicronS. The collection of the blood samples was conducted along with the swab and saliva samples, which were used for the RT-qPCR-based detection of COVID-19 by the hospital and further confirmed by us as mentioned earlier [41]. The processing of the samples was carried out in a NABL-certified (MC-3486) and ICMRapproved diagnostic laboratory for COVID-19 testing following regulatory guidelines and protocols as described previously [41]. Blood samples were processed to obtain the plasma by centrifuging at 2000× g for 15 m, and the processed samples were either used immediately or stored at −80 • C for future use.

RT-qPCR and PCR Analysis
Saliva samples from patients were collected and processed as previously described [41].
To check for the Omicron variant, a PCR assay was performed with S gene target failure (SGTF), as per the published report [42]. The IL6 and IFNB RT-qPCR were performed using the RNA extracted (QIAGEN RNA extraction kit) from NHBE cells. The RNA was subjected to cDNA synthesis (QIAGEN RNA extraction kit) followed by PCR using KAPA SYBR FAST qPCR master mix (KAPA). The assay was performed using Rotor-Gene Q (QIAGEN).
The mitochondrial DNA copy number was determined in the patient plasma samples using 50 µL as the starting sample. DNA was extracted in 20 µL of TE buffer (Vivantis Technologies, Malaysia). Finally, 2 µL of DNA was used per sample of PCR reaction containing the primers spanning various regions of the mitochondrial genome as shown in Supplementary Table. PCR was performed using 2X PCR Master Mix (Thermo Fisher Scientific, Boston, MA, USA) or amaR OnePCR (GeneDireX, Inc., Vegas, NV, USA). The mtDNA copy number was determined as previously described [43]. Similarly, the mtDNA copy number was determined in the chromatin-immunoprecipitated plasma samples (described later in this section). The amplicon of the representative extracted mtDNA (ND2) was also confirmed by performing gel-electrophoresis as described by us previously [41].

Cell Culture and Transfection
A549 cells (adenocarcinoma human alveolar basal epithelial cells) were obtained from ATCC and cultured in Ham's F12 medium (Sigma) supplemented with 10% heatinactivated FBS (Gibco), 2 mM L-glutamine (Invitrogen), and 100 U/mL penicillinstreptomycin (Invitrogen). Cells were maintained in a 5% CO 2 incubator at 37 • C. Transfection was carried out using lipofectamine 3000 (Thermo Fisher Scientific, Boston, MA, USA) according to the manufacturer's protocols. Briefly, 30,000 cells were seeded in 24-well plates on the coverslip for immunofluorescence. After 48 h at 70% confluency, the cells were transfected with the respective plasmid's master mix, which included 250 µL of reduced serum medium Opti-MEM (Thermo Fisher Scientific, Boston, MA, USA), 1 µg plasmid, and 2 µg transfection reagent. Cells were incubated for 6 h in the transfection mix, followed by the removal of the transfection complexes and the addition of fresh media. After 24 h, the cells were processed for immunoblotting or immunofluorescence. Normal Human Bronchial Epithelial (NHBE) cells (Lonza Bioscience) were maintained in a 5% CO 2 incubator at 37 • C. The cells were grown in BEGM™-2 BulletKit medium (Lonza Bioscience). The medium was changed every 48 h.

Human Mesenchymal Stem Cell Isolation, Culture, and Characterization
Human bone marrow samples were obtained from healthy individuals with no history of infectious disease, diabetes, or metabolic syndrome at the Apollo Indraprastha Hospital, New Delhi, after obtaining proper patient consent. MSCs were isolated from the collected bone marrow as described by us previously [35]. Stem cell marker analyses were performed using flow cytometry. Briefly, cells were fixed with 4% paraformaldehyde (PFA) (Sigma) and incubated in a blocking solution (1.5% normal goat serum and 3% BSA) for 30 min, followed by staining with the respective primary fluorophore-conjugated antibodies for 1 h. CD44-FITC, CD73-PE, and CD90-PE were used as positive markers, and CD45-FITC was used as a negative marker. Analysis was performed by acquiring 10,000 events, and the samples were run in triplicate using a BD FACSMelody instrument. FlowJo™ v.10 software (BD Biosciences, East Rutherford, NJ, USA) was used for analysis.

DNA Constructs and Plasmids
All SARS-CoV-2 protein-encoding constructs were obtained as kind gifts from Dr. Nevan Krogan [44]. An empty vector was generated by deleting the open reading frame of NSP4 using the restriction enzymes EcoRI and BamH1 (NEB) while leaving the Strep II tag intact. BAK, BAX, and MCL1 shRNA were generated using the pLKO.1 vector backbone (Addgene#269279). The scrambled sequence was replaced with respective shRNA sequences as shown in Supplementary Table. The shRNA sequences for BAK and BAX were obtained as synthetic fragments; PCR was amplified to generate the double strands and then cloned into the vectors using the NdeI and EcoRI (NEB) restriction sites. The sequences were generated using the online (Custom Dicer-Substrate siRNA (DsiRNA)) tool from Integrated DNA Technologies. MCL1 was obtained as a double-stranded gene fragment with the corresponding NdeI and EcoRI restriction sites and introduced into the pLKO.1 vector. Three different shRNA target sequences were tested for BAK and BAX and two for MCL1.
The lentiviral-encoded MCL1 vector was obtained as a kind gift from Scott J Dixon [45]. The Bcl-xL construct used in this study was generated using pLYS1-FLAG-MitoGFP-HA (Addgene#50057) and the ORF of the Bcl-xL plasmid (Addgene #140749) obtained from Scott J Dixon. The ORF of the Bcl-xL was obtained by PCR amplification with a forward primer containing Nhe1 and a reverse primer containing the BsrGI sites. Both the PCRamplified products and the pLYS1-FLAG-MitoGFP-HA vector (referred to as mitoGFP) were digested followed by ligation. The final vector obtained was named pLV-Bcl-xL after sequence confirmation by restriction and sequencing. Similarly, lentiviral-encoded PINK1 was generated by PCR amplification of the PINK1 ORF from the pEYFP-N1-Pink1 (Addgene #101874) with the Nhe1 and BsrGI (NEB) sequences in the primers. The amplified product was inserted into the mitoGFP. The final plasmid lentiviral-encoded PINK1 plasmid was confirmed by sequencing and named pLV-PINK1.

Preparation of Lentiviral Particles
To produce lentiviruses, the plasmid-encoding gene of interest (transfer plasmid) was co-transfected with the packaging vector plasmid (pCMV-dR8.2 dvpr; Addgene #8455) and envelope-encoding plasmid (pCMV-VSV-G; Addgene # 8454) in HEK 293T cells at a ratio of 4:4:2, respectively (total 10 µg of DNA), in T25 flasks. Cells were transfected using 20 µL of LTX (Invitrogen) and 5 ml of complete DMEM media (Gibco). After 48 h of transfection, the media containing the lentiviral particles was collected and centrifuged at 500× g for 10min. Then, the clarified supernatant was incubated with Lenti-X (Lenti-X™ Concentrator, Takara, SHG, Japan) in a 3:1 ratio. This supernatant-LentiX mixture was incubated for 48 h at 4 • C. After 48 h, the mixture was centrifuged at 1500× g for 45 min at 4 • C. The concentrated pellet was resuspended in 100 µL of complete media and used immediately at MOI of 5 or stored at −80 • C as aliquots. The MOI was calculated by real-time quantitative PCR using a lentivirus titration kit (Lenti-X RT-qPCR Titration Kit, Takara, SHG, Japan). All transduction experiments in A549 cells with various SARS-CoV-2encoded plasmids or plasmids encoding shRNA for BAK, BAX, and MCL1 were carried out at MOI 5 for 48 h or 72 h unless specified. Similarly, for the generation of genetically engineered MSC shBAK+MCL1 , cells were transduced with the respective lentiviral particles for 48 h before the co-culture experiment. To evaluate the expression of BAK or MCL1 in MSCs, the cells were fixed, followed by immunofluorescence with the respective antibodies.

SYTOX GREEN Assay
Flow cytometry analysis was performed to measure cell death by staining the cells with SYTOX Green Nucleic Acid Stain (Thermo Fisher Scientific, Boston, MA, USA). The assay was performed according to the manufacturer's instructions. Briefly, cells were stained with the dye for 20 min and washed before acquisition. About 10,000 cells were acquired per sample and the samples were run in triplicate. Flow cytometry analysis was performed using a BD Accuri C6 Plus, and the data were analyzed using FlowJo™ v.10 Software.

TUNEL Assay
TUNEL assays were performed by us as described previously and following the manufacturer's instructions (Promega) [35]. NHBE cells were first stained with CellTracker Green (CTG) before co-culture for identification. After 48 h, the media were replaced with fresh media containing SARS-CoV-2 viruses at MOI of 0.2. After 48 h of transduction, the virus-containing medium was removed and the cells were washed with 1× PBS four times and fixed with 4% PFA. The fixed cells were used for further immunofluorescence analysis. All infections were performed in the BSL3 facility at the Regional Centre for Biotechnology, Faridabad, Haryana.

SARS-CoV-2 Infection Model
A549 cells were transduced with lentiviral particles derived from the ACE2/TMPRSS2expressing vector (Addgene #154987), and cells stably expressing ACE2/TMPRSS2 were selected using puromycin selection. Around 30,000 ACE2/TMPRSS2-expressing A549 cells were seeded on coverslips in 24-well plates. After 48 h, the media were replaced with fresh media containing SARS-CoV-2 viruses at MOI of 0.2. After 48 h of transduction, the viruscontaining media were removed and the cells were washed with 1× PBS four times then fixed with 4% PFA. The fixed cells were used further to perform immunofluorescence. All the infections were carried out in the BSL3 facility at the Regional Centre for Biotechnology, Faridabad, Haryana.

Flow Cytometry Analysis
As described by us previously [46], the mitochondrial reactive oxygen species (mtROS) were measured by staining the cells with MitoSOX Red (Thermo Fisher Scientific, Boston, MA, USA) and the mitochondrial membrane potential (∆Ψm) was measured with TMRE (Sigma-Aldrich, St. Louis, MO, USA); 50,000 cells/well were seeded in 24-well plates. After 24 h of seeding, the cells were transfected with different plasmid constructs in triplicate and allowed to grow for another 24 h. The cells were then washed using 1× PBS, stained with MitoSOX or TMRE as per the recommended concentrations, and incubated for 15 min, followed by washing and analysis by flow cytometry using a BD Accuri C6 plus or BD Melody, respectively. Similarly, cells were stained with Mito-Tracker Green for 20 min and analyzed using a BD Accuri C6. Ten thousand events were acquired for each experiment and the data were analyzed using FlowJo™ v.10 Software.

Drug Treatments
To induce mitochondrial DNA release from A549 cells, 30,000 cells were seeded in 24-well plates with coverslips. After 48 h of seeding, cells were transduced/transfected with mitoGFP lentivirus/respective plasmids. After 24 h of transduction/transfection, one set of cells was treated with a combination of three different drugs, namely, ABT737 (10 µM, Calbiochem, CA, USA), Actinomycin D (1 µM, Calbiochem), and zVAD (20 µM, Sigma-Aldrich, St. Louis, Mo, USA) for 2 h and the other set of the cells was used as a control. The treated cells were processed for immunofluorescence.
Similarly, cells were treated with Antimycin A at a concentration of 100 nM (Sigma) for 24 h to induce mtROS and FCCP at 10 µM for 4 h to induce a decrease in ∆Ψm. For the co-culture experiments, A549 cells were treated with rotenone at 100 nM for 24 h before co-culture with MSCs.

Transmission Electron Microscopy
TEM imaging was performed to investigate the ultrastructural changes in the mitochondria. Briefly, lentivirus-transduced A549 cells were fixed in 4% PFA and 2.5% glutaraldehyde. Excess fixative was removed using a 0.1M sodium cacodylate buffer and processed for block preparation in 2% agar. Another round of fixation was conducted in 2% osmium tetra oxide for 1 hr. Dehydration of the samples was carried out using 30%, 50%, 70%, and 100% ethanol, then mounted upon resin and polymerized at 60 • C for 72 h. The ultramicrotome (Leica EM UC7, Leica, Wetzlar, Germany) was used for cutting the ultrathin sample sections (63 nm) placed on the copper grids and stained with 5% uranyl acetate and 0.2% lead citrate. The analysis of the chopped sections was carried out in a 200 KVA transmission electron microscope (Tecnai G2 20 twin, FEI, Hillsboro, OR, USA). The condition of the mitochondrial cristae was analyzed manually, and mitochondria with no clear double membrane and imperfect cristae were labeled as disrupted.

Chromatin Immunoprecipitation
The conditioned media were collected from cultured A549 cells transduced with ORF9b and NSP4 or empty vector lentivirus particles. These were concentrated using 10 kDa Amicon filters. For each sample, 20 mL of the media was collected and processed. The concentrated media were then used for chromatin immunoprecipitation. An amount of 5 µL anti-TFAM (CST) antibody was added to 500 µL media supernatant and incubated overnight at 4 • C on a rotatory shaker. The following day, 50 µL of Protein A/G Sepharose (Abcam, Boston, MA, USA) was added and incubated for 4 h at RT. The beads were then spun down and the supernatant was discarded. The bead pellets were then washed with a 50 µL lysis buffer twice. Half of the samples were processed for immunoblotting (described later) and the other half were used for PCR (as mentioned above).

Immunofluorescence
Cells seeded on coverslips with a confluency of 60-70% were used for immunofluorescence according to pre-established protocols [46]. In brief, the 4% PFA fixed cells were washed with 1X PBS for 10 min, then permeabilized in a blocking solution containing 1X PBS with 0.1% Triton X-100 and 5% goat serum for 1 h at RT. Cells were incubated with primary antibodies (2 µg/mL) overnight in 1X PBS containing 0.01% Triton X-100 and 2% goat serum at 4 • C. The next day, the cells were washed thrice with 1x PBS and incubated with fluorophore-tagged secondary antibodies (1 µg/mL) in a buffer (same as the primary antibody) for 1 h at RT. Cells were washed five times with 1X PBS for 5 min each at RT after secondary antibody incubation. Finally, the coverslip was mounted on a frosted slide with DAPI-containing mounting media.

Preparation of Mitochondrial and Cytosolic Fractions
Mitochondrial and cytosolic (mito-cyto) fractions were separated using a mitochondrial isolation kit (Sigma) according to the manufacturer's instructions. Briefly, cells from two T75 flasks were harvested when they reached a confluence of 70-80% after a PBS wash, followed by trypsinization. Cells were pelleted at 1200 rpm for 5 min. Pellets were then resuspended in 400 µL of extraction buffer mixed with a protease inhibitor cocktail (Sigma) at a ratio of 1:100 and kept for 20 m on ice. To lyse the cells, each sample was passed 15-20 times through a 26-gauge syringe. Cells were analyzed for 50% lysis by live cell count using trypan blue. This was followed by centrifugation at 600× g for 10 min at 4 • C. The supernatant was then processed further by high-speed centrifugation at 11,000× g for 10 min at 4 • C. The obtained supernatant was the cytoplasmic fraction, whereas the pellet constituted the mitochondrial fraction. The extracted mitochondrial fractions were resuspended in 50 µL cell lysate mixed with a protease inhibitor cocktail at a 1:100 ratio and quantified. The cytosolic fractions were used for probing with an anti-TFAM antibody (1:1000, Abcam, Boston, MA, USA). For immunoblotting of TFAM in the cell supernatant, the sample was subjected to filtration through a 0.2 µm filter followed by concentration with 10 kDa Amicon filters. Protein estimation was conducted and 30 µg of the concentrated supernatant protein was used and resolved in 10% gel.

Immunoblotting
For the immunoblotting experiments, cells were cultured in the appropriate media and transduced with the respective lentiviral-encoded plasmids or drugs. For total cell lysate (TCL) preparation, cells were washed once with PBS, followed by trypsinization and cell lysis using a RIPA buffer (Thermo Fisher Scientific, Boston, MA, USA) mixed with a protease inhibitor cocktail (Sigma) and DTT 0.1M (Sigma) at a ratio of 100:1:1. For cell lysis, the mixture was vortexed for 30 min (15 s each at 5 min intervals) at 4 • C. Isolated proteins were then quantified using the BCA method. An amount of 30 µg of protein was resolved in a 10% SDS-PAGE gel (unless mentioned otherwise elsewhere), then transferred on a PVDF membrane. After blocking of the PVDF membrane in 5% BSA in PBS, the respective primary antibodies were used (Figure 1n Figure S4: anti-cGAS, anti-NLRP3, and anti-GAPDH) for overnight primary antibody incubation at 4 • C. All the primary antibodies were used in 1:1000 dilutions in PBST/TBST + 0.1% Tween20 along with 5% BSA. The full blots are shown in Figure S9.
After overnight incubation, the blots were washed thrice with PBST/TBST and probed with the corresponding secondary antibodies (1:10,000) conjugated with HRP and incubated for 1 h at RT. After washing the blots with PBST/TBST, they were incubated with chemiluminescence reagent (Invitrogen/Thermo Fisher Scientific, Boston, MA, USA) for 30 s-5 min according to band intensity and visualized using a ChemiDoc instrument (Bio-Rad Laboratories, Hercules, CA, USA). The image signals were quantified by performing image analysis using the Image J software.
Chromatin immunoprecipitated samples for TFAM immunoblotting were prepared as described above under the heading "Chromatin immunoprecipitation". An amount of 30 µg of the extracted protein was used. Blots were incubated with anti-TFAM antibody overnight at 4 • C. Anti-rabbit secondary conjugated with HRP was used at a dilution of 1:10,000 and incubated for 1 h at RT. For anti-GFP immunoblotting, the cell supernatant was collected from scrambled or MCL1 shRNA-transduced cells after 48 h. Similarly, supernatants from NSP4 and ORF9b or VEC-transduced cells were collected after 48 h. The supernatant was subjected to filtration through 0.2 µm filters, then concentrated using 10 kDa Amicon filters, followed by protein estimation. An amount of 20 µg of protein was used and resolved in 10% gel. Anti-rabbit GFP and E-Cadherin antibodies were together used overnight at 4 • C. Anti-rabbit secondary conjugated with HRP was used at a dilution of 1:10,000 and incubated for 1 h at RT. Blots were developed using chemiluminescence substrate (Invitrogen/Thermo Fisher Scientific, Boston, MA, USA) at different exposure times on a ChemiDoc Imaging System (Bio-Rad Laboratories, Hercules, CA, USA).

Immunoprecipitation
Immunoprecipitation of Strep-II-tagged NSP4 and ORF9b was performed using 500 µg of total cell lysate with 5 µg of anti-Strep-II antibody (Abcam, Boston, MA, USA) or rabbit anti-IgG (Abcam, Boston, MA, USA) as a control. The cell lysate-antibody mix was incubated for 12 h at 4 • C. Immunoprecipitation was performed as mentioned in the commercially available immunoprecipitation kit (Abcam, Boston, MA, USA). Briefly, Protein A/G Sepharose (Abcam, Boston, MA, USA) blocked with BSA-skimmed milk (2:3 ratio) was added to the lysate and incubated for 1 h at room temperature. The antibodyprotein lysate complex was eluted in 40 µL of 2X SDS-PAGE loading buffer and boiled for 5 min, followed by centrifugation to collect the eluent. SDS-PAGE was performed and the respective antibodies (anti-BAK, anti-BAK, anti-MCL1, and anti-Bcl-xL) were used. The VeriBlot Detection Reagent conjugated with HRP was used to probe the blots.

Co-Culture Studies and Intercellular Mitochondrial Transfer Assay
For the co-culture experiments, A549 or NHBE cells were transduced with NSP4 and ORF9b lentiviral-encoded particles for 24 h or with rotenone as previously described [41]. Simultaneously, MSCs were transduced with mitoGFP lentiviral particles. For the coculture experiments, A549 cells were stained with Cell Tracker Deep Red (CTDR; Thermo Fisher Scientific, Boston, MA, USA) for 20 min and washed thrice with PBS. After this, MSC and A549 were co-cultured at a 1:1 ratio for another 24 h. Following co-culture, cells were visualized under live conditions. Images were obtained using a Nikon confocal Ti2E or Zeiss microscope integrated with an Apotome 2 (ZEISS Axio Observer 7). Mitochondrial donation from MSCs to A549 or NHBE was determined by specifically examining the mitoGFP signal in CTDR-positive cells and analyzed using Image J software. Similarly, to check the expression of BAK or MCL1 in A549 cells after co-culture, cells were stained with the respective antibodies. The signal from CTDR-positive A549 cells was quantitated, which represented the expression of BAK or MCL1 in these cells.

Pro-Inflammatory Cytokine, Cytochrome c in NHBE Cells
Patient plasma samples were used to measure the different cytokines by ELISA. The assay was performed in the hospital according to the manufacturer's protocols. The cytokines profiled in the patient plasma samples included IL-6 (Krishgen Biosystems, Mumbai, MH, India), TNF-α (Krishgen Biosystems, Mumbai, MH, India), and IL-10 (Krishgen Biosystems, Mumbai, MH, India). The ferritin assay was performed according to the manufacturer's protocols (Calbiotech, El Cajon, CA, USA). The human cytochrome c immunoassay was performed according to the manufacturer's instructions (R&D Systems, Minneapolis, Min). Briefly, 20 µL of patient plasma sample was used for the assay and the input volume was adjusted to 100 µL using an assay diluent. The OD was measured at 450 nm using a microplate reader (Thermo Fisher Scientific, Boston, MA, USA).
To measure the pro-inflammatory response in NHBE cells, total levels of IL6 (BioLegend, San Diego, CA, USA) and IFN-γ (BioLegend, San Diego, CA, USA) were measured in total cell lysates. ELISA was performed according to the manufacturer's protocols. Briefly, 10 µg of the total cell lysate was used for the measurement. The readings were obtained by the spectrophotometer (Multiskan SkyHigh Microplate Spectrophotometer, Thermo Fisher Scientific, Boston, MA, USA). Calculations were performed as previously described [35].  Figure 5h, the images were acquired in cell culture media without phenol red. Similarly, for the live-cell co-culture imaging ( Figure 7 and Figure S7), the images were acquired in media without phenol red. The images shown in Figure S8a were acquired with the EVOS FLoid Imaging System (Thermo Fisher Scientific, Boston, MA, USA) using a 20 × objective. All other images shown were acquired with a TCS SP8 LSCM (Leica) using a 63 × oil objective with 1.4 NA.

Image Acquisition and Analysis
Post-image acquisition and image processing were performed using Image J software. Mitochondrial shape classification was performed following the previously described protocols [47]. For the integrated density calculation, the image was first processed to mask the signal outside by specifically focusing on the mitochondrial stain, which was stained with either TOM20 or MitoTracker Red. The signal outside the mitochondria was eliminated after processing in all channels. Next, images were split into three different channels as previously mentioned [46]. To specifically measure the signal associated with the mitochondria, only channel images with corresponding signals of the TFAM signal or PicoGreen or anti-DNA stained (representing mtDNA) images were used. The channel with the mtDNA signal and mitochondrial signals was selected and the threshold was set independently using the OTSU threshold method. Finally, the mask of the segmented image of the mitochondria was generated and applied to the TFAM segmented image, thus clearing the signal outside of the mitochondria. Finally, the image was redirected to measure the integrated density parameters. Similarly, for the calculation of the BAX and BAK puncta (Figure 4i,k), the signal outside the mitochondria was masked and the number of puncta associated with the mitochondria was calculated. The number of puncta was calculated per image and represented as puncta per cell.
For colocalization analysis, the 'EzColocalization' plugin [48] was used, as previously described [46]. The Mander's overlap coefficient (or Mander's coefficient) was calculated between BAK and NSP4, and MCL1 and ORF9b, as shown in Figure 6g,h. The Mander's coefficient measures the percentage colocalization as described by us previously [46].

Statistical Analyses
P-values for single comparisons were calculated using an unpaired t-test, whereas one-way ANOVA with Tukey's post hoc test was used for independent multiple comparisons using GraphPad Prism software (Prism v.9). All the experiments were repeated a minimum of three times, and the figures shown in this manuscript were prepared using GraphPad Prism.

Positive Correlation between mtDNA Release and Pro-Inflammatory Immune Response in Patients with COVID-19
This study aimed to establish the correlation between mtDNA release and the SARS-CoV-2-induced pro-inflammatory responses in patients with COVID-19. A set of mitochondrial genes was examined to identify a marker that can consistently and reliably detect mtDNA in the cell-free (cf) plasma samples of patients. The initial screening of 10 randomly selected patient plasma samples revealed that ND2 provided the most consistent results compared with the other tested mitochondrial genes (Figure 1a). Using ND2 as an mtDNA marker, the plasma samples of 89 SARS-CoV-2-positive cases, 35 SARS-CoV-2-negative cases, 14 convalescent cases (recovered; previously infected cases confirmed to be SARS-CoV-2-negative using quantitative real-time polymerase chain reaction (RT-qPCR)), and 24 new cases suspected to be infected with the Omicron variant were profiled ( Figure S1a). Compared with those in the SARS-CoV-2-negative and COVID-19-recovered (hereafter referred to as recovered) groups, the mtDNA copy number was significantly higher in the SARS-CoV-2-positive group (Figure 1b). The mtDNA levels in patients suspected to be infected with the Omicron variant (samples collected between January 2022 and March 2022; hereafter referred to as OmicronS) were lower than those in patients infected with other variants (samples collected between June 2020 and October 2020; hereafter referred to as COVID- 19), which are reported to yield S gene-positive results [41].
To examine the correlation between the cf-mtDNA levels and pro-inflammatory response, the levels of cytokine storm-associated key inflammatory markers were analyzed using an enzyme-linked immunosorbent assay. The IL-6, IL-1β, TNF-α, and IL-10 levels were consistently upregulated in patients with COVID-19 (Figure 1c-f). Similar to the mtDNA levels, the levels of PICs (IL-6, IL-1β, and TNF-α) and IL-10 were consistently downregulated in patients with OmicronS infection. Correlation analysis revealed that the IL-1β, IL-6, TNF-α, and ferritin levels were positively correlated with the mtDNA levels ( Figure 1g-l). These results suggest that the plasma mtDNA levels are closely correlated with the PIC response in patients with COVID-19.
To examine if cf-mtDNA directly elicits the inflammatory response, an in vitro model of normal human bronchial epithelial (NHBE) cells was used. The cells were incubated directly with the plasma samples of the SARS-CoV-2-positive (serologically positive) and SARS-CoV-2-negative (serologically negative) cases [41]. Cells were incubated with plasma for 24 h(h) and total RNA was extracted and subjected to RT-qPCR analysis. The IL1B mRNA levels were significantly upregulated in the cells incubated with the plasma of the SARS-CoV-2-positive cases ( Figure S1b). Compared with that of the SARS-CoV-2negative cases, the plasma of the SARS-CoV-2-positive cases induced cell death at 48 h post-incubation ( Figure S1c,d).
The plasma of patients with COVID-19 comprises various pro-inflammatory factors and DAMPs hence, the ability of mtDNA to induce the pro-inflammatory response was examined. Chromatin immunoprecipitation (ChIP) of TFAM (a mitochondrial DNA nucleoid-associated protein) was performed using 10 pooled patient samples divided into eight groups each ( Figure 1m). The immunoprecipitated mtDNA was subjected to DNA extraction. The presence of ND2 in the extracted DNA was verified using PCR (Figures 1n and S1e). Total protein and mRNA were extracted from NHBE cells incubated with various copy numbers of the extracted mtDNA. Consistent with the results obtained using the SARS-CoV-2-positive plasma samples ( Figure S1B), purified mtDNA from these samples dose-dependently upregulated the IL1B and IL6 mRNA levels in NHBE cells (Figures 1p and S1f). Additionally, purified mtDNA upregulated the IL-1β and IL-6 levels (Figures 1q and S1g). Consistent with the data shown in Figure S1c,d, mtDNA isolated from the SARS-CoV-2-positive plasma samples increased NHBE cell death ( Figure S1h,i). These findings indicate that mtDNA isolated from the plasma of patients with COVID-19 induces pro-inflammatory responses and cell death.  . All data are represented as mean ± SEM of three independent experiments, analyzed by one-way ANOVA (b,p,q) and unpaired t-tests (c-f) using Graphpad Prism software. ** p < 0.01; *** p < 0.001; **** p < 0.0001; ns: not significant.
Next, this study aimed to determine if mtDNA release is a general phenomenon associated with SARS-CoV-2 infection-induced mitochondrial damage. As mitochondrial damage is often associated with the release of pro-apoptotic Cytochrome c (Cyt c) [49], the plasma levels of Cyt c were measured in patients with COVID-19. Consistent with the findings of mtDNA release, the plasma Cyt c levels in the SARS-CoV-2-positive cases were upregulated compared with those in the SARS-CoV-2-negative cases ( Figure S1j). These findings indicated that cf-mtDNA is a robust indicator of mitochondrial damage in patients with COVID-19 and that it can be used as a potential biomarker along with other pro-inflammatory markers to determine COVID-19 severity.

SARS-CoV-2 Infection Induces Epithelial Cell Mitochondrial Dysfunction and mtDNA Release
To evaluate the effect of SARS-CoV-2 on the morphology and function of the mitochondria in airway epithelial cells, SARS-CoV-2 infection-permissible A549 cells were generated by stably expressing ACE2/TMPRSS2 and infected with SARS-CoV-2 (strain USA-WA1/2020) for 48h ( Figure 2a). Next, the cells were fixed and immunostained with an anti-spike protein antibody to confirm SARS-CoV-2 infection ( Figure 2b). The mitochondrial structural changes in the infected cells were examined using immunofluorescence analysis with the anti-TOM20 (OMM protein) antibody. SARS-CoV-2 infection induced robust mitochondrial structural changes. The infected cells exhibited punctate/blob mitochondrial morphology, whereas the control cells exhibited an elongated tubular network of mitochondria ( Figure 2c). These changes were also confirmed using image analysis (Figure 2d). High-resolution LIGHTNING microscopy imaging analysis revealed swollen cristae and disrupted inner mitochondrial membranes (IMM; represented as damaged mitochondria; 8.85 ± 1.63 (control group) vs. 43.9 ± 1.63 (SARS-CoV-2-infected group)) (Figures 2e,f, and S2a) and decreased mitochondrial mass in the SARS-CoV-2-positive samples ( Figure 2g). To investigate the functional impact of these mitochondrial structural changes, the mitochondrial pro-apoptotic Cyt c levels were determined using immunostaining. The Cyt c signal associated with the mitochondria was markedly downregulated (269 ± 8 (control group) vs. 140.1 ± 4.7 (SARS-CoV-2-infected group)) in the SARS-CoV-2 infected cells, indicating a robust release of Cyt c into the cytosol (Figure 2h,i). Next, the effect of SARS-CoV-2 infection on pro-inflammatory mtDNA release was examined using LIGHTNING microscopy with the anti-TFAM (to detect mtDNA) and anti-TOM20 (to visualize OMM) antibodies. In the control group, TFAM was associated with the mitochondria, especially within the mitochondria. However, the mitochondria were largely devoid of mtDNA with mtDNA visualized just outside the mitochondria in the SARS-CoV-2-infected cells, indicating the release of mtDNA (Figures 2j and S2b). As a positive control for the mtDNA release experiments, cells were treated for 2h with a combination of actinomycin D, ABT-737, and zVD, which are known to induce mtDNA release. The TFAM signal intensity was significantly downregulated in the SARS-CoV-2-infected cells (259.3 ± 10.8, 124.7 ± 3.15, and 88.7 ± 3.2 in the control, SARS-CoV-2-infected, and positive control groups, respectively) ( Figure 2k). We hypothesized that the extrusion of mtDNA into the cytoplasm may have activated the cytosolic DNA sensors, such as the cGAS-STING1 and/or NLRP3 inflammasome. However, the expression levels of cGAS or NLRP3 were not significantly different between the control and SARS-CoV-2-infected groups ( Figure S2c-e). These findings indicate that SARS-CoV-2 infection induces robust mitochondrial structural changes and mtDNA release in lung epithelial cells (Figure 2l), which were consistent with the mtDNA release observed in patients with COVID-19. In contrast to the findings of previous studies on endothelial cells, mtDNA extruded from airway epithelial cells did not induce the expression of canonical DNA sensors in the cytosol in this study [50]. LIGHTNING microscopy images showing the colocalization of anti-TFAM antibody-stained mtDNA (magenta) within the anti-TOM20 antibody-stained mitochondria (green). Insets show that in the control group, mtDNA appears within the mitochondria. In the SARS-CoV-2-infected group, most mitochondria are devoid of mtDNA (red arrowheads) and in some instances, the mtDNA is localized outside the mitochondria (yellow arrowheads). (k) Quantitative analysis of the downregulation of mtDNA associated with the mitochondria in SARS-CoV-2-infected cells from images shown in (j) (n = 13).

SARS-CoV-2 Proteins NSP4 and ORF9b Induce Mitochondrial Dysfunction and mtDNA Release
The presence of mtDNA in the circulation is attributed to robust mitochondrial damage extending from OMM to IMM. To determine the SARS-CoV-2 proteins involved in mitochondrial dysfunction and mtDNA release, the effects of all 29 SARS-CoV-2 proteins on mtDNA release were examined using immunofluorescence analysis. A549 cells were individually transfected with plasmids encoding viral proteins. The release of mtDNA in the transfected cells was examined using immunofluorescence analysis with the anti-TFAM (to label mtDNA) and anti-TOM20 (to label mitochondria) antibodies to detect the levels of mtDNA associated with the mitochondria. Compared with that in the empty vector (VEC)transfected cells, the amount of mtDNA associated with the mitochondria was significantly lower in cells transfected with NSP4, ORF6, and ORF9b (Figure 3a). The combination of actinomycin D, ABT-737, and zVD) was used as a positive control. Transfection with NSP2, ORF6, and ORF9b decreased the mtDNA signal intensity associated with the mitochondria (Figure 3b). Similarly, the levels of Cyt c associated with the mitochondria were examined using immunofluorescence. Transfection with NSP4 and ORF9b significantly decreased the Cyt c signal associated with the mitochondria (Figure 3i,j). Next, the following two major mitochondrial functional parameters were measured using flow cytometry: mitochondrial reactive oxygen species (mtROS) and mitochondrial membrane potential (∆Ψm). Transfection with NSP2, NSP4, NSP6, NSP8, ORF3a, ORF6, and ORF9b upregulated the mtROS levels and significantly downregulated ∆Ψm. Antimycin A and carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) were used as the positive controls for the mtROS and ∆Ψm experiments, respectively ( Figure S3b,c). To increase the efficiency of ectopic expression, lentiviral particles encoding the selected SARS-CoV-2 proteins with a marked adverse effect on the mitochondria were generated. The effect of these lentiviral particles was evaluated individually or in combination with mtDNA and the mitochondrial function parameters. Cells transduced with NSP4, ORF6, or ORF9b exhibited the most significant downregulation of the mtDNA signals associated with the mitochondria, Cyt c signals associated with the mitochondria (Figure 3c), and ∆Ψm ( Figure 3d) and upregulation of mtROS levels (Figure 3e). A similar trend was observed in the transduced cells stained with PicoGreen, which has been used previously to measure mtDNA release [24] ( Figure S3d). These results indicate that compared with other SARS-CoV-2-encoded proteins, NSP4, ORF6, and ORF9b have the most prominent effect in inducing mitochondrial dysfunction and mtDNA release in airway epithelial cells (Figure 3f). Next, the combinatorial effect of NSP4, ORF6, and ORF9b on mtDNA release was evaluated. Transduction with the combination of NSP4 and ORF9b (hereafter referred to as N4 + 9b) resulted in the most marked downregulation of mtDNA signals associated with the mitochondria (Figure 3g); mtDNA appeared to be present outside the mitochondria in some of the transduced cells (Figure 3h). Additionally, transduction of cells with N4 + 9b resulted in the robust downregulation of the Cyt c signals associated with the mitochondria (Figure 3i,j). Immunoblotting of the cytosolic fraction revealed the appearance of the released Cyt c in the cytoplasm (Figure 3k). Based on these findings, NSP4 and ORF9b were selected for further experiments.     ActD+ABT+zVD   100   150   200   I1 I2 I3 I4 I5 I6 I7 I8 I9 I10 I11 I12 I13 I14 I15 I16 I17 I18    . Data are plotted as percentage of cell death. All data are represented as mean ± standard error of mean from three independent experiments. Statistical analyses (one-way analysis of variance (b,c,h,n,p) and unpaired t-tests (j,q,r,t)) were performed using Graphpad Prism software. * p < 0.05, ** p < 0.01; *** p < 0.001; **** p < 0.0001; ns: not significant. Scale bar: 10 µm.

NSP4 and ORF9b Induce Pro-Inflammatory mtDNA Extrusion from the Cells
To investigate the effect of the released mtDNA on the induction of cytosolic DNA sensors and inflammasome formation, the expression levels of cGAS and NLRP3 were examined. Consistent with the results obtained in the SARS-CoV-2 infection experiments, the expression of these sensor proteins was not induced ( Figure S4). Based on these results, the levels of mtDNA in the cytoplasm and culture supernatant were examined. The expression of TFAM (mtDNA marker) was localized to the cytosol in the N4 + 9b-transduced cells. However, the cytosolic TFAM level in the N4 + 9b-transduced cells was significantly lower than that in positive control-treated cells (Figure 3k,m). The mtDNA levels in the culture supernatant of the N4 + 9b-transduced cells were higher than those in the culture supernatant of the positive control-treated cells (Figure 3l,m). These results explain the failure of released mtDNA to induce cytosolic pathogen recognition receptors (PRRs) as most of the mtDNA appears to be secreted outside the cells.
To further examine if the TFAM observed in Figure 3i is associated with mtDNA released from the mitochondria, the culture supernatant of N4 + 9b-transduced cells was subjected to TFAM ChIP analysis (as described in Figure 1m), followed by immunoblot-ting analysis (Figure 3n). DNA was isolated from the immunoprecipitated complex and subjected to RT-qPCR. The number of mtDNA copies was significantly upregulated in the culture medium of the N4 + 9b-transduced cells (Figure 3o,p). This is consistent with the increased mtDNA release observed in patients with COVID-19 (Figure 1b). To examine the immunogenic response of the released mtDNA, NHBE cells were treated with various copy numbers of isolated mtDNA and the expression of IL1B was examined. mtDNA from N4 + 9b-transduced cells dose-dependently upregulated the IL1B mRNA ( Figure 3q) and total IL-1β (Figure 3r) levels. Further, mtDNA-induced cell death in NHBE cells was examined using flow cytometric analysis with Sytox green. Treatment with mtDNA from the N4 + 9b-transduced cells significantly increased cell death (Figure 3s,t). These findings indicate that NSP4 and ORF9b induce the release of pro-inflammatory mtDNA, which subsequently induces the expression of pro-inflammatory IL-1β and cell death in the surrounding cells.

NSP4 and ORF9b Induce Mitochondrial Damage and BAX/BAK-Dependent Mitochondrial Macropore Formation
As SARS-CoV-2-infected cells exhibited robust mitochondrial structural changes and decreased mitochondrial mass (Figure 2c-g), the effect of transduction with N4 + 9b on mitochondrial morphology and mitochondrial mass was examined. N4 + 9b-transduced A549 cells were stained with MitoTracker Red (MTR) and MitoTracker Green (MTG) to analyze the mitochondrial shape and mass, respectively. At 48 h post-transduction, confocal microscopy and flow cytometry analyses were performed. Compared with the healthy elongated network's mitochondrial morphology, in the VEC-transduced cells, robust changes in mitochondrial morphology, especially increased numbers of punctate and blob mitochondria, were observed in the N4 + 9b-transduced cells. Rotenone was used as a positive control for the experiments to evaluate the changes in mitochondrial shape ( Figure S5a). Image analysis confirmed the presence of a decreased mitochondrial network in the N4 + 9b-transduced cells (Figure 4a). To visualize the changes in mitochondrial shape at a higher resolution, the transduced cells stained with the anti-TOM20 antibodies were subjected to LIGHTNING microscopy. Consistent with the previous results, the mitochondria mostly exhibited a round/donut-shaped morphology with extensive inner membrane damage (Figure 4b). The appearance of round-shaped mitochondria was accompanied by a decreased mitochondrial mass (Figure 4c).
Mitochondrial structural changes with deceased ∆Ψm are often associated with mitochondrial clearance through mitophagy [51]. To determine if the decreased ∆Ψm in the N4 + 9b-transduced cells was due to increased mitophagy, the expression levels of PINK1 and PARKIN in the mitochondrial fraction and the mitochondrial mass (an indicator of mitochondrial turnover) were examined. However, PINK1 accumulation and Parkin translocation to the mitochondria were not significantly affected in the N4 + 9b-transduced cells (Figure 4d). Next, a lentiviral construct of PINK1 was generated to evaluate the effect of PINK1 overexpression on mitophagy. Mitochondrial mass markedly declined in the PINK1-expressing N4 + 9b-transduced A549 cells owing to the induction of mitophagy (Figure 4e). However, the cells could not survive beyond 72 h in the culture, suggesting that PINK1 overexpression-induced mitophagy promotes cell death rather than cell survival ( Figure S5b,c). Thus, PINK1 overexpression in the N4 + 9b-transduced A549 cells led to excessive mitophagy, which may not be therapeutically viable. Mitochondrial damage is often associated with mitochondrial permeability transition pore (mPTP) opening. To investigate the opening of mPTP in the N4 + 9b-transduced cells, an mPTP assay was performed using the fluorescent probe calcein-AM. Ionomycin was used as a positive control. The calcein-AM intensity in the N4 + 9b-transduced cells was significantly lower than that in the VEC-transduced cells (Figure 4f,g). mPTP is the result of mitochondrial inner membrane permeabilization (MIMP) and mitochondrial outer membrane permeabilization (MOMP). Recent studies have demonstrated that BAX and BAK-mediated MOMP facilitates mtDNA release and is preceded by mPTP formation [24,25]. Furthermore, BAX/BAK pore formation and the subsequent mtDNA release activate the pro-inflammatory response in the cell [25]. Thus, the stabilization of mitochondrial BAK and the translocation of cytosolic BAX were examined in the N4 + 9b-transduced cells using confocal microscopy. The mitochondrial levels of BAK and BAX were upregulated in N4 + 9b-transduced cells. A similar expression pattern was observed in cells treated with macropore formation inducers (Figures 4h-k and S5d,e). The upregulated BAX and BAK levels on the mitochondrial surface were further confirmed by subjecting the mitochondrial extracts to immunoblotting analysis (Figure 4l). Apoptotic stimuli promote the association between BAX and BAK on the mitochondria and consequently induce BAX/BAK macropore formation [25]. Thus, the formation of similar structures was examined in the N4 + 9b-transduced cells. BAK and BAX puncta colocalized with the mitochondrial TFAM, suggesting the formation of macropores as characterized earlier [25] (Figure 4m). The prerequisite for macropore formation is the Bcl-xL and MCL1-mediated inhibition of the apoptotic pathway. Thus, the expression levels of Bcl-xL and MCL1 (anti-apoptotic proteins) were examined. The MCL1 levels were significantly downregulated in the mitochondrial fractions, whereas the Bcl-xL levels were not affected in the N4 + 9b-transduced cells, suggesting that MCL1 is involved in macropore formation (Figure 4n).
To understand whether the mitochondrial translocation of BAX, stabilization of BAK, and MCL1 downregulation were prerequisites for mtDNA release, the expressions of BAK and BAX were knocked down using short hairpin RNA (shRNA) (shBAK and shBAX, respectively) (Figure 4o,p), and MCL1 (anti-apoptotic protein involved in mtDNA release) was upregulated using the overexpression vector [25,52] (Figure 4q). Simultaneous BAK knockdown and MCL1 overexpression markedly decreased mtDNA release from the mitochondria compared with other conditions (Figure 4r). This indicates that MCL1 has an essential role in mtDNA release along with BAX/BAK-induced macropores.

MCL1 Induces Inner Mitochondrial Membrane Vesicle Formation
The critical involvement of MCL1 in mtDNA release encouraged us to explore the roles of MCL1 in processes other than those involved in promoting BAX/BAK macropore formation. The effect of MCL1 knockdown ( Figure S6a) on IMM/mitochondrial cristae was examined ( Figure S5a). Electron microscopy (EM) imaging revealed that MCL1 knockdown resulted in extensive IMM/cristae disruption (Figure 5a,b). The mitochondrial matrix comprised vesicular structures, which were observed in most of the imaged damaged mitochondria. Next, the inner membrane stability was examined using LIGHTNING microscopy. Cells were stained with anti-TOM20 antibodies and co-transfected with an inner membrane-targeted mitochondrial green fluorescent protein (GFP) (IMM GFP ; inner membrane reporter). MCL1 knockdown downregulated the IMM GFP signal, suggesting inner membrane rupture (Figure 5c,d). The IMM GFP signal was also downregulated in the healthy tubular network of the mitochondria. Additionally, the IMM GFP signal in the N4 + 9b-transduced cells was lower than that in the VEC-transduced cells ( Figure S6b). However, BAX/BAK shRNA-transfected groups did not exhibit marked downregulation in the IMM GFP signal (Figure 5d). Cells were transfected with IMM GFP and probed with the anti-DNA and anti-TOM20 antibodies to simultaneously visualize both the inner and outer membranes and mtDNA. mtDNA was always co-associated with the inner membrane-like vesicles (Figure 5e,f). The association of mtDNA with the inner membrane was further confirmed by the colocalization of IMM GFP with TFAM in both the N4 + 9b-transduced and MCL1 knockdown cells (Figure 5g). Next, cells co-transfected with IMM GFP and Scarlet-TFAM were subjected to live-cell imaging, which revealed small vesicular structures enclosing the mtDNA (Figure 5h). A similar degree of mtDNA association with COX-IV (inner membrane protein) was observed in the N4 + 9b-transduced and shMCL1-transduced groups ( Figure S6c,d). To further evaluate the impact of these inner membrane vesicular structures, the expression of IMM GFP was examined in the culture supernatant. The expression of IMM GFP was detected in the culture supernatant of MCL1 knockdown and the N4 + 9b-transduced cells (Figure 5i). These results suggest that NSP4 and ORF9b mitigate the MCL1-mediated suppression of inner membrane destabilization and mtDNA release into the inner membrane vesicles (Figure 5j) in addition to facilitating BAX/BAK macropore formation on the OMM.

NSP4 Binds to BAK and ORF9b Interacts with MCL1 to Regulate mtDNA Release
The above-mentioned results indicate that NSP4 and ORF9b regulate macropore formation and inner membrane vesicle formation through BAX/BAK and MCL1. To elucidate the direct interaction of NSP4 and/or ORF9b with the key proteins inducing MOMP, A549 cells were individually transfected with NSP4 or ORF9b containing a Strep-II tag. The total cell lysate was prepared and subjected to NSP4 and ORF9b co-immunoprecipitation using the anti-Strep-II antibodies, followed by immunoblotting with anti-BAX, anti-BAK, anti-MCL1, and anti-Bcl-xL antibodies. BAK co-immunoprecipitated with NSP4, whereas MCL1 co-immunoprecipitated with ORF9b (Figure 6a-d). These results were further confirmed using immunofluorescence and colocalization analyses, which revealed enhanced BAK signals and the colocalization of BAK with NSP4 (Figure 6e,g). The overall MCL1 signal was downregulated with the remaining signal colocalizing with ORF9b, which was consistent with the immunoprecipitation results (Figure 6f,h). Additionally, the interaction of the NSP4 protein (N4 T492I ) found in the recently identified Omicron variant of SARS-CoV-2 with BAK was examined. However, no significant changes were observed in the association of BAK with N4 T492I (Figure 6e,g). Similarly, the mtDNA release was not significantly different between the N4 T492I -transduced cells and wild-type NSP4-transduced cells (Figure 6i). NSP4 transduction significantly increased the BAK levels on the mitochondrial surface (Figure 6j,k). In contrast, ORF9b transduction significantly downregulated MCL1 expression as evidenced by the decreased signal associated with mitochondria (Figure 6l,m). Furthermore, the expression of MCL1 associated with IMM was examined. Transduction with ORF9b downregulated IMM-associated MCL1 (Figure 6n). These results suggest that the NSP4 and ORF9b of SARS-CoV-2 promote mitochondrial macropore formation and that MCL1 regulates inner membrane stability and vesicle formation. NSP4 recruits BAK to the mitochondria by directly interacting with it to form the BAX/BAK macropore (Figure 4m). Transduction of ORF9b mitigated the inhibitory effect of MCL1 on BAX/BAK macropore formation and inner membrane stability most likely by hindering the MCL1 turnover. Statistical analyses (unpaired t-tests (b,f) and one-way analysis of variance (d)) were performed using Graphpad Prism software. * p < 0.05; **** p < 0.0001; ns: not significant (non-parametric t-test). Scale bars: 0.5 (a), 0.2 (ROI), or 10 µm (c,e,g,h).

BAK Knockdown and MCL1-Overexpressing MSCs Rescue Cell Death through Functional IMT
Previously, we demonstrated that IMT rescues epithelial cell death through the donation of healthy mitochondria via TNTs [35]. To evaluate if the donated healthy mitochondria can rescue mtDNA release and cell death, A549 cells were transduced with VEC (A549 VEC ) or N4 + 9b (A549 N4+9b ) for 72 h. The transduced A549 cells were then stained with Cell-Tracker Red (CTDR) immediately before co-culturing with MSCs expressing mito-GFP. After 24 h of co-culture, the GFP signal was detected in A549 N4+9b cells (indicated by the presence of both mito-GFP and CTDR (yellow)) (Figure 7a), suggesting mitochondrial uptake from MSCs. Image analysis revealed a significant increase in the mitochondrial transfer from the MSCs to A549 N4+9b cells (Figure 7b). Analysis of mito-GFP signals in A549 cells revealed that TNFAIP2 (which regulates TNT-mediated mitochondrial transfer) knockdown significantly inhibited IMT (Figure 7b). This indicated that mitochondrial donation from MSCs to A549 cells was dependent on TNTs. To examine whether IMT could rescue cell death, unlabeled MSCs were co-cultured with CTDR-stained A549 cells. After 24 h of co-culture, cells were stained with Sytox green and subjected to flow cytometry analysis. Compared with that in the A549 VEC -MSC co-culture, cell death was not markedly rescued in the A549 N4+9b -MSC co-culture (Figure 7c). Next, the effect of IMT on the metabolic function of recipient A549 N4+9b cells was examined. The donated mitochondria were not sufficient to decrease the mtROS levels or attenuate mitochondrial depolarization in recipient cells (Figure 7d). These results encouraged us to generate genetically engineered MSCs that can markedly rescue the metabolic defect and attenuate cell death in A549 N4+9b cells. MSCs were genetically engineered by targeting BAK and MCL1 (Figure 7e). BAK was knocked down (Figure S7a,b) and MCL1 was overexpressed ( Figure S7c,d) in MSCs using a similar approach to the A549 cells (Figure 4o,q). These engineered MSCs (MSC shBAK+MCL1 ) retained the stem cell markers similar to the control cells (MSC Con ). This suggests that the simultaneous BAK knockdown and MCL1 overexpression did not affect the stem cell phenotype of MSCs ( Figure S7e). MSC shBAK+MCL1 co-cultured with A549 N4+9b cells exhib-ited enhanced mitochondrial donation to the recipient cells (Figure 7f). Representative images of IMT by MSC shBAK+MCL1 are shown in Figure S7f. The donated mitochondria exhibited healthy tubular morphology (Figure 7g), attenuated mtROS, and restored ∆Ψm in A549 N4+9b cells (Figure 7h). Similarly, MSC shBAK+MCL1 restored mitochondrial function in NHBE cells by donating functional mitochondria (Figure 7i). Notably, the BAK expression was consistently lower in A549 N4+9b and MCL1 levels were increased after the co-culture with MSC shBAK+MCL1 (Figure 7g,h).   (l,m). However, instead of staining the cells with the anti-TOM20 antibody, the cells were stained with the anti-COX-IV antibody to locate the MCL1 signal associated with the mitochondrial inner membrane. All data are represented as mean ± standard error of mean from three independent experiments. Statistical analyses (one-way analysis of variance (g,h,i) and unpaired t-tests (k,m,n)) were performed using Graphpad Prism software. * p < 0.05, ** p < 0.01; *** p < 0.001; **** p < 0.0001; ns: not significant (non-parametric t-test). Scale bars: 20 (e,f) or 10 µm (j,l).
To validate the IMT potential of MSC shBAK+MCL1 in a more holistic model, SARS-CoV-2 infected cells were used (as shown in Figure 2). MSC shBAK+MCL1 exhibited increased mitochondrial donation to the SARS-CoV-2-infected A549 cells (Figure 7j,k). IMT from MSC shBAK+MCL1 restored healthy mitochondrial morphology in SARS-CoV-2-infected cells (Figure 7l). These results suggest that MSC shBAK+MCL1 retain their functional IMT potential and rescue SARS-CoV-2-mediated mitochondrial dysfunction in the recipient airway epithelial cells.

MSC shBAK+MCL1 Attenuate Inflammatory mtDNA Release from Airway Epithelial Cells and Rescue Cell Death
Next, the effects of the donated mitochondria on the apoptotic response and proinflammatory mtDNA release were examined. The co-cultures of MSCs (unlabeled) and SARS-CoV-2-infected A549 (CTDR-stained) cells were subjected to Cyt c immunostaining. The intracellular Cyt c levels in CTDR-stained SARS-CoV-2-infected A549 cells (indicating decreased Cyt c release) co-cultured with MSC shBAK+MCL1 were significantly upregulated compared with those in CTDR-stained SARS-CoV-2-infected A549 cells co-cultured with the non-targeted scrambled control (Srm) and VEC2 (overexpression empty vector for MCL1)transfected MSCs (MSC Srm+VEC2 ) (Figure 8a,b). Next, the effect of MSC shBAK+MCL1 on mtDNA release from SARS-CoV-2-infected A549 cells was examined. Upon co-culture with MSC shBAK+MCL1 , the levels of mtDNA associated with the mitochondria were significantly upregulated in the recipient A549 N4+9b cells, suggesting decreased mtDNA release from the mitochondria (Figure 8c,d). MSC shBAK+MCL1 mitigated A549 cell death, which was also verified using the terminal deoxynucleotidyl transferase dUTP nick-end labeling assay (TUNEL) (Figure S8a,b). To verify these findings in the A549 N4+9b cells, MSC shBAK+MCL1 were co-cultured with A549 N4+9b . Compared with SARS-CoV-2 infection (Figure 8a,b), the MSC shBAK+MCL1 co-culture decreased Cyt c release (Figure 8e,f), mitigated mtDNA release (Figure 8g,h), and markedly decreased cell death ( Figure S8c) in the A549 N4+9b cells. Similar therapeutic effects of MSC shBAK+MCL1 were observed upon quantification of mtDNA copies in the recipient A549 N4+9b cells in co-culture ( Figure S8d).   To explore whether the therapeutic effect of MSC shBAK+MCL1 involves the mitigation of the pro-inflammatory effect on the A549 N4+9b cells, the pro-inflammatory and antiapoptotic effects of the co-culture supernatant were examined ( Figure 8i). As shown in Figure 8j,k, the co-culture supernatant downregulated the IL1B mRNA and the secreted IL-1β levels in NHBE cells. Additionally, the A549 N4+9b -MSC shBAX+MCL1 co-culture supernatant attenuated NHBE cell death compared with the A549 N4+9b -MSC Srm+VEC2 co-culture supernatant (Figure 8l). Thus, these results provide the first comprehensive evidence that the modulation of mitochondrial pore formation by targeting BAK activation and enhancing anti-apoptotic MCL1 expression potentiates the therapeutic efficacy of MSCs. Additionally, the donation of functional mitochondria from human MSC shBAK+MCL1 prevented macropore formation in the recipient SARS-CoV-2-infected or NSP4/ORF9b-expressing airway epithelial cells, which can be a potential therapeutic strategy for COVID-19. Based on their enhanced therapeutic efficacy, we named these genetically modified cells IMAT-MSCs (intercellular mitochondrial transfer-assisted therapeutic MSCs).

Discussion
SARS-CoV-2-induced mitochondrial damage is an emerging pathological determinant in COVID-19 [50,[53][54][55][56][57]. Extensive mitochondrial ultrastructural changes and functional impairments are primarily observed in infected airway epithelial cells [31,58] in addition to endothelial cells [50], monocytes [54,57], and T cells [55,59]. Under some conditions, mitochondrial damage is accompanied by pathogenic mtDNA release. Several studies have demonstrated that the circulating mtDNA levels in patients with COVID-19 are positively correlated with disease severity [12,13]. The findings of this study further demonstrated the value of circulating mtDNA as a potential biomarker to predict COVID-19 severity and the pro-inflammatory response. In this study, mtDNA extracted from patients with COVID-19 elicited a pro-inflammatory response and promoted cell death in primary human airway epithelial cells. These novel findings encouraged us to elucidate the molecular mechanism of SARS-CoV-2-induced mtDNA release and develop a rational stem cell-based approach with potential clinical applications for the treatment of other diseases in addition to COVID-19.
The extrusion of mtDNA from the mitochondria is reported in various inflammatory and infectious diseases [10,11]. However, the molecular mechanism underlying mtDNA release under pathological conditions, especially SARS-CoV-2 infection, has not yet been elucidated. This study, for the first time, elucidated the molecular pathway involved in SARS-CoV-2 infection-induced mtDNA release. Mechanistically, the SARS-CoV-2-encoded proteins NSP4 and ORF9b synergistically induce mtDNA release. NSP4 stabilizes BAK on the outer membrane through direct interaction. Subsequently, BAK recruits BAX to induce outer membrane macropore formation. These BAX/BAK macropores were previously reported to be formed upon exposure to apoptotic stimuli, usually under caspase inhibition conditions [24,25]. Similarly, infections, such as severe fever with thrombocytopenia syndrome virus infections, can induce BAX/BAK-dependent mtDNA release [26]. This is the first study to demonstrate that SARS-CoV-2 mediates the formation of these macropores in target airway epithelial cells. Other have also shown an intricate relationship between SARS-CoV-2 and mitochondrial function. Notably, a comprehensive study from the Madesh lab has elucidated the molecular basis of SARS-CoV-2-mediated autophagy in cardiomyocytes [22]. The authors have generated a comprehensive map of the SARS-CoV-2-encoded proteins interacting with various mitochondrial resident proteins and thereby inducing mitochondrial fragmentation, calcium imbalance, and cell death. Flynn et al. performed a screening using an integrated system of clustered regularly interspaced palindrome repeat-caspase (CRISPR-Cas) and the comprehensive identification of RNA-binding proteins by mass spectrometry and reported the localization of SARS-CoV-2 RNA with mitochondrial proteins. In particular, MRM2 was highly enriched with the SARS-CoV-2 RNA fractions [60]. Other studies have also reported the mitochondrial localization of SARS-CoV-2 RNA [23], which was proposed to be mediated by TOM20. These findings suggest that SARS-CoV-2 infection is associated with extensive mitochondrial dysfunction.
In addition to outer membrane macropore formation, mtDNA release from the mitochondrial matrix is dependent on overcoming the inner membrane barrier. MIMP was speculated to be a passive event that is initiated after macropore formation. This theory postulated that the inner membrane ruptures upon exposure to the cytosol after outer membrane macropore formation, leading to the release of mtDNA into the cytosol [24,25]. This study provided the first molecular evidence that mtDNA release from the inner membrane is not a passive event and that it is a well-coordinated event regulated by MCL1. EM, immunofluorescence, and biochemical analyses revealed that MCL1 inhibition induces inner membrane vesicle formation. Genetically overexpressing MCL1 prevents inner membrane damage and mtDNA release. The effect of MCL1 on inner membrane vesicle formation was further validated in cells expressing ORF9b. ORF9b induced inner membrane vesicle formation in transfected airway epithelial cells by directly interacting with MCL1, which prevented the accumulation of MCL1. The distinct roles of MCL1 have been previously reported. MCL1 localized to the outer membrane suppresses the apoptotic signal, whereas that localized to the matrix aids in the maintenance of inner membrane integrity and the cristae ultrastructure and prevents mPTP formation [52,61]. However, this study proposed a novel mechanism through which MCL1 regulates inner membrane vesicle formation and subsequent mtDNA packaging into these vesicles. Mitochondrion-derived vesicles were previously characterized as comprising only the outer membrane or both inner and outer membranes [62]. However, the findings of this study suggest that vesicles enclosing mtDNA are predominantly derived from the inner membrane and are subsequently extruded via BAX/BAK macropores on the outer membrane. A comprehensive characterization of these vesicles and their excision mechanism will provide useful insights into this phenomenon, which must be explored in future studies.
The release of mtDNA into the cytosol or the extracellular space is intricately associated with the activation of different PRRs [63,64]. Upon release into the cytosol, mtDNA activates the cGAS-STING1 pathway, which activates the type I interferon response [29,30]. The activation of the cGAS-STING pathway may be beneficial during viral infections and potentiate the anti-viral activity of immune cells [29,30] or result in pathological consequences in conditions, such as autoimmune diseases [65]. In SARS-CoV-2 infection, mtDNA released from endothelial cells activates the pro-inflammatory response through the cGAS-STING pathway [50]. However, the findings of this study indicate that in the airway epithelial cells, mtDNA release is not coupled with intrinsic cGAS-STING activation or NLRP3 inflammasome formation. This study demonstrated that a large pool of released mtDNA is secreted extracellularly via inner membrane vesicles. Thus, mtDNA extruded from the mitochondria is protected from cytosolic DNA sensors, leading to the suppression of an immediate anti-viral interferon response. This may explain the impaired interferon response reported in patients with COVID-19 [32,66]. The presence of a minor percentage of mtDNA released without encapsulation in IMM vesicles cannot be ruled out, although this minor population may not be adequate to activate cytosolic DNA sensors. These effects of mtDNA release and the underlying mechanisms may be cell-type-specific. SARS-CoV-2-infected endothelial cells are reported to activate the cGAS/STING pathway [50]. The immediate activation of the NLRP3 pathway in the infected cells can also be ruled out. This supports the notion that the SARS-CoV-2-infected cells do not undergo immediate cell death, which may otherwise limit the propagation of the virus in the target cells.
The extracellularly released mtDNA can exert its effect on the neighboring cells by activating the TLR9 signaling pathway [13,67,68], which can trigger downstream signaling pathways, leading to the induction of PICs and chemokines. cf-mtDNA derived from the plasma of patients with COVID-19 induced a robust pro-inflammatory response and subsequently promoted cell death in primary airway epithelial cells. Furthermore, mtDNA extracted from N4 + 9b-transduced cells exerted similar pro-inflammatory and pro-apoptotic effects on airway epithelial cells. Based on these findings, we propose a new model of mtDNA release during SARS-CoV-2 infection. According to this model, the mtDNA avoids immediate exposure to cytosolic DNA sensors, leading to the inhibition of the protective anti-viral signaling of host cells. During mtDNA packaging into the inner membrane vesicles, a robust systemic pro-inflammatory response is elicited. Consistent with the findings of this study, previous studies [12,13] have reported a strong positive correlation between cf-mtDNA release and pro-inflammatory molecules in clinical serum samples of patients with COVID-19. Thus, the kinetics of these events must be evaluated to establish a correlation between the release of mtDNA (both from the mitochondria and extracellularly) and cell death during SARS-CoV-2 infection in a time-dependent manner. The effect of extracellularly released mtDNA on bystander cells and a similar phenomenon during infection by other viruses must be investigated in the future. Further studies are needed to explore whether the canonical TLR9 signaling or other pathways are activated by the extracellular mtDNA.
To devise a novel approach for therapeutic intervention based on the findings of this study, a clinically approved approach was utilized to mitigate SARS-CoV-2-induced mitochondrial damage and mtDNA release. According to the strategy employed in this study, MSCs were engineered to become resilient to macropore formation and inner membrane vesicle excision. Based on the results in airway epithelial cells, BAK was knocked down and MCL1 was overexpressed in MSCs using the lentiviral-based gene delivery system. Upon co-culture with SARS-CoV-2-infected or N4 + 9b-transduced cells, genetically engineered MSCs (IMAT-MSCs) exhibited enhanced therapeutic potential by mitigating mitochondrial damage and pro-inflammatory mtDNA release in the recipient cells. Mechanistic studies revealed that IMAT-MSCs donated functional mitochondria to the recipient cells via TNTs. Previously, we had reported that TNTs act as molecular connecting bridges for IMT between MSCs and airway epithelial cells [35]. This finding has also been reported in different cell types by other studies [40]. The IMT ability of MSCs is a relatively new paradigm that explains the beneficial effects of MSC transplantation. Recent studies have demonstrated that the phenomenon of IMT is conserved across different cell types both in vitro and in vivo, which provides an opportunity to devise new-generation targeted stem cell-based therapies. Our mechanistic study of the therapeutic effect of MSCs was confined to an in vitro model system which sometimes may not recapitulate the in vivo disease condition and the physiological response. Thus, to strengthen these findings, the therapeutic effect of these engineered MSCs should be evaluated in appropriate in vivo models.
MSCs have exhibited enormous translational potential for the treatment of various respiratory conditions. Successful clinical outcomes of MSC transplantation in H7N9induced acute respiratory distress syndrome [69,70] and other respiratory diseases [70] have encouraged the therapeutic application of these cells for COVID-19, which is associated with similar complications. Currently, more than 100 clinical trials evaluating the efficacy of MSCs for the treatment of COVID-19 are enlisted (ClinicalTrials.gov). Recently, several groups have published clinical trial results evaluating the safety and efficacy of MSCs for COVID-19. The results of phase 1 clinical trials, which were published in early 2020, demonstrated the safety of MSC transplantation and reported that the mortality rate was significantly low among patients with COVID-19 receiving MSC transplantation [71] (and reviewed in [72]). Consistently, phase II trial results were also promising [73] with a recent 1-year follow-up showing long-term benefits [72]. These clinical results indicate that MSCs will provide a viable long-term alternative therapeutic option to decrease COVID-19-associated mortality. Therefore, the genetic modulation of these cells (as demonstrated in this study) to develop next-generation MSC therapy will have applications for treating other diseases in addition to COVID-19.
In summary, the findings of this study have immediate clinical applications, especially considering the ability of new emerging SARS-CoV-2 variants to evade vaccinemediated immunity. Based on the beneficial effects of MSCs, the Food and Drug Administration (FDA) has recently approved the first MSC (modulated to express TACR1)-based therapy for COVID-19.

Limitations of Study and Future Directions
The major limitation of this study was that the safety and efficacy of IMAT-MSCs were not evaluated in preclinical or clinical settings. However, similar lentiviral-based cellular therapies are reported to be safe and have yielded promising clinical outcomes. One such approach is the chimeric antigen receptor (CAR) T-cell therapy. The currently FDA-approved CAR T-cell products are generated using a similar third-generation lentiviral system to deliver the CAR genes into the target cells. Thus, we believe that IMAT-MSCs should not have safety concerns. However, comprehensive preclinical and clinical assessments are warranted, which is beyond the scope of this study. The other limitation of this study was the limited availability of SARS-CoV-2 samples for detailed molecular studies and the limited number of Omicron variant samples for evaluation of mtDNA levels owing to the low number of hospitalizations during the third wave (January 2022 to March 2022) of COVID-19 in India. Further, most of the study has been conducted using an imaging-based approach such as immunofluorescence with a selected cell population. We used multiple replicates and the experiments were performed more than three times to avoid any bias in image selection and quantitation. Even then, biochemical assays and other sensitive assays, such as flow cytometry where a large population of cells can be evaluated, will complement some of our experimental results.  (20) and  patients, as measured by ELISA. Data is represented in nanograms/mL. All data are represented as mean ± SEM of three independent experiments, analyzed by non-parametric t test using Graphpad Prism software. *** p < 0.001; **** p < 0.0001; ns: not significant. Figure Figure 3b. Data is represented as mean ± SEM of three independent experiments. Scale bar: 10 µm. Figure S4: NSP4 and ORF9b of SARS-CoV-2 does not induce expression of cytosolic DNA sensors. Representative immunoblot showing cGAS and NLRP3 expression in A549 cells transduced with N4+9b for 48h, followed by cell lysis and total protein extraction. Anti-cGAS and anti-NLRP3 antibodies were used to detect cGAS and NLRP3 respectively. A combination of ActD+ABT+zVD (2h) was used as positive control to induce the expression of these DNA sensors. Figure S5: NSP4 and ORF9b lead to mitochondrial structural changes and macropore formation. (a) Representative images showing mitochondrial shape changes in cells transduced with VEC or N4+9b, as observed with mitotracker red (MTR; pseudocolored as green) with images analysis shown in Figure 4a. Rotenone (Rot) was used as a positive control to induce mitochondrial shape changes. (b) Representative flow cytometry histograms showing cell death, as measured by SYTOX green staining. A549 cells were co-transduced with N4+9b with or without PINK1 lentiviral expression vector (PINK1pLV). VEC1 represents the backbone construct for N4 and 9b plasmids, while VEC2 represents the backbone construct for PINK1 plasmid. Numbers indicate cell death percentage. (c) Cell death represented as % cell death analyzed from the histograms shown in panel b (n = 6). (d) Representative images showing BAK and (e) BAX (red) associated with anti-TOM20 stained mitochondria (green) in A549 cells transduced either with VEC or N4+9b. A combination of ActD+ABT+zVD (2h) was used as positive control. These images represent all the panels of images shown in Figure 4h,j. Data is represented as mean ± SEM of three independent experiments, analyzed by non-parametric t test, * p < 0.05; *** p < 0.001. Scale bars: 10 µm. Figure S6: MCL1 downregulation and NSP4/ORF9b induce inner membrane disruption. (a) RT-qPCR data showing relative fold change expression of MCL1 upon transduction of A549 cells with Scramble (Srm) or shRNA targeting MCL1, followed by RNA extraction and RT-qPCR analysis (n = 6). (b) Representative images showing rupture of inner mitochondrial membrane of A549 cells transduced with N4+9b, stained with inner membrane targeted GFP (IMMGFP) and anti-TOM20 to probe outer membrane. (c) Representative images showing of COX-IV expression in cells transduced with N4+9b, or (d) MCL1 shRNA, and stained with anti-TOM20. All data are represented as mean ± SEM of three independent experiments, analyzed by non-parametric t test, **** p < 0.0001. Scale bars: 10 µm.  Figure 7f. (g,h) Quantitive analysis of the images by specifically measuring the fluorescence signal in A549 cells after co-culture with MSCSrm+V2 or MSC-shBAK+MCL1. The data represents the signal of BAK on mitochondria (g) or MCL1 on mitochondria (h). A549 cells were transduced with vector or N4+9b before co-culture. All data are represented as mean ± SEM of three independent experiments, analyzed by non-parametric t test, **** p < 0.0001. Scale bar: 20 µm. Figure  . All data are represented as mean ± SEM of three independent experiments, analyzed by non-parametric t test, *** p < 0.001; **** p < 0.0001. Scale bar: 100 µm. Figure S9: Source data of all the blots. The original full blots are shown for the representative figures shown in Main and Supplementary [74,75].  P1/12-21.12.2020) were also obtained from Jamia Millia Islamia as mentioned by us previously. Patient consent was obtained while collecting the samples according to the ICMR GCP guidelines.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
There is no data associated with the manuscript as source data for the repository.